What a small retailer should measure after a price change
I raised prices once and sales dropped sharply. The obvious theory was that the new price was too high. I could not prove it. Buyers were already stretched by inflation, which made the price a convenient suspect and a hard one to isolate. The drop was real. The cause was a guess.
Most small retailers reach this moment with a few dozen orders a week and no analytics team. The goal is not a perfect experiment. It is to avoid a confident reaction to noise. Here is what to record and calculate, in order.
Write down the change before you look at results
On the day the price moves, record five things in a spreadsheet:
- Which products or variants changed, the old price, and the new price.
- The date and time the new price went live on every channel you sell on.
- Anything else that changed that week: ad budget, email sends, discount codes, shipping rates, stock levels, product photos, site changes.
- Your expectation. For example: “We expect units to fall about 10 percent and margin per order to rise.”
- The date you will review it.
The expectation matters more than it looks. Without it, any outcome gets explained afterward as if you had predicted it.
The five numbers to track
Revenue alone hides what happened. Split it:
Revenue = sessions × conversion rate × average order value
Each term answers a different question.
| Number | How to calculate it | What it tells you |
|---|---|---|
| Sessions by source | Visits from search, paid, email, social, marketplace | Whether the audience changed |
| Conversion rate | Orders ÷ sessions | Whether visitors still buy at the new price |
| Average order value | Revenue ÷ orders | Whether basket size moved |
| Contribution margin per order | Price minus product cost, shipping, fees, and ad cost | Whether you earn more or less per sale |
| Return and refund rate | Returned orders ÷ orders | Whether the price changed who is buying |
A price change acts most directly on conversion rate. If conversion held steady and sessions fell, look at traffic, not price.
A worked example (illustrative numbers)
Four weeks before the change: 2,000 sessions, 60 orders, $60 average order value. That is a 3.0 percent conversion rate and $3,600 in revenue.
Four weeks after: 1,500 sessions, 42 orders, the same $60 average order value. That is a 2.8 percent conversion rate and $2,520 in revenue.
Revenue fell 30 percent. It is tempting to blame the price. But sessions fell 25 percent, while conversion fell only from 3.0 to 2.8 percent, a relative drop of about 7 percent. Most of the loss came from traffic. Maybe a campaign ended or a search ranking slipped. The price may still have cost some sales, but the data points to a smaller share of the problem than the headline number.
Margin can rise while revenue falls
Before calling a price change a failure, check the break-even volume. With a unit cost C, an old price P0 and a new price P1:
Maximum volume loss before profit falls = 1 − (P0 − C) ÷ (P1 − C)
Say an item costs $20 to make and ship, and the price moves from $40 to $44. Margin per unit goes from $20 to $24. You can lose up to 1 − 20/24, about 17 percent of units, and still earn the same total profit. Revenue would drop in that case. Profit would not.
Use contribution margin, not gross margin, if you can. Marketplace fees and per-order ad cost change the answer.
Pick three time windows
A single before-and-after comparison is too crude. Use three:
- Days 1 to 7: the reaction. Expect noise. Email subscribers and returning customers may behave differently from new visitors. Items already in carts, or orders placed from saved links, may still carry the old price. Do not decide anything here.
- Weeks 2 to 4: the settled read. This is your main comparison. Compare the same weekdays. A Tuesday and a Saturday are different days.
- Weeks 6 to 8: the lagging effects. Check return rates, repeat orders, customer questions about price, and reviews. If you make items to order, remember orders take days to ship, so measure by order date, not ship date.
Before reading anything, ask how many orders you have per window. Order counts vary by roughly the square root of their size even when nothing changes. At 20 orders a week, a swing of 4 or 5 orders in either direction is ordinary. If your windows hold fewer than about 30 orders, treat a result as a hint and extend the window rather than drawing a conclusion.
Separate the price from the calendar
Seasonality is the most common false explanation, in both directions. Two simple checks help.
Same period last year. If sales for the same weeks last year rose 15 percent over the previous month, a flat result this year may mean the price did cost you something. Without last year’s data, use the shape of your own history: compare the same weeks in the months before, and note holidays.
A control group. Compare the products whose price changed with similar products whose price did not, over the same dates. If both groups fell by the same amount, the price is not the main story. If only the changed group fell, the evidence is stronger. A small shop can do this by hand, with only a few products on each side.
Gift products show why this matters. A collection such as personalized bracelets for mothers is bought around dates, not at an even pace through the year, so a price change made just before a gifting date can look better than it was, and one made just after can look worse. For any gift line, compare against the same weeks of the previous year, not the weeks just before the change.
Check what else moved
Before assigning cause, run down this list for the comparison period:
- Traffic mix. Did paid, organic, email and marketplace traffic keep the same shares? Cheap, low-intent paid traffic can lower conversion with no change in price.
- Stock. Was a best seller sold out or a size missing for part of the period?
- Discounts. A removed promo code raises the price customers actually pay more than the list price change does.
- Shipping and delivery promise. A higher shipping fee or a longer delivery estimate can change the decision at checkout.
- Measurement. Did analytics, checkout or tagging change? A broken tracking tag can look exactly like a sales drop.
Mistakes that cost the most
- Judging after three days. The first week mostly measures noise and carry-over.
- Reading store-wide revenue. A change on 10 products disappears in the total, and a traffic dip elsewhere can get blamed on the price.
- Changing price again too soon. Two changes in one month leave you with no clean reading from either.
- Using different definitions. If “orders” counted refunded orders before and not after, the comparison is broken.
- Ignoring margin. A change that loses 8 percent of units can still be a good one.
- Skipping the record. Memory keeps the sales drop and loses the conditions around it.
Keep a short decision log
After the review, write four lines: what changed, what you measured, what you concluded, and how sure you are. “Conversion on changed products fell from 3.0 to 2.8 percent with traffic mix stable. Weak evidence of a price effect. Not enough orders to be sure” is a fine entry. It is better than a confident one you cannot support.
The next price change then starts from a record, not a mood. After my own drop, the most useful thing I could have had was a page showing what I had measured and what I had only guessed.
About the author
Aviad Faruz is CEO of FARUZO, a made-to-order personalized jewelry business. He manages its ecommerce pricing, marketplace operations and product data.